def plotAllPages(fh):
# tol = 5 # This is the tolerance for histogram rounding
fig_list = []
ax1_list = []
pdf = PDFDocument(fh)
print "Created by: %s" % pdf.get_creator()
# print SelectedPDF
# flt = 'LTTextLineHorizontal'
# flt = ['LTPage','LTTextLineHorizontal']
flt = ["LTPage", "LTFigure", "LTLine", "LTRect", "LTImage", "LTTextLineHorizontal", "LTCurve"]
# flt = ['LTPage','LTChar']
for i, page in enumerate(pdf.get_pages()):
# page = next(doc.get_pages())
# layout = page.layout() # LTPage
box_list = LeafList().populate(page, interested=flt)
ModalHeight = pdftables.calculate_modal_height(box_list)
diagnostic_data = pdftables.TableDiagnosticData(box_list, {}, {}, [], [])
fig, ax1 = plotpage(diagnostic_data)
fig_list.append(fig)
ax1_list.append(ax1)
title = "page %d" % (i + 1)
fig.suptitle(title)
# print "Page %d" % (i+1), ElementCount
print box_list.count()
print "Modal character height: %d" % ModalHeight
return fig_list, ax1_list
def get_tables(fh):
"""
Return a list of 'tables' from the given file handle, where a table is a
list of rows, and a row is a list of strings.
"""
pdf = PDFDocument(fh)
return get_tables_from_document(pdf)
def plotAllPages(fh):
#tol = 5 # This is the tolerance for histogram rounding
fig_list = []
ax1_list = []
pdf = PDFDocument(fh)
print "Created by: %s" % pdf.get_creator()
# print SelectedPDF
#flt = 'LTTextLineHorizontal'
#flt = ['LTPage','LTTextLineHorizontal']
flt = [
'LTPage', 'LTFigure', 'LTLine', 'LTRect', 'LTImage',
'LTTextLineHorizontal', 'LTCurve'
]
# flt = ['LTPage','LTChar']
for i, page in enumerate(pdf.get_pages()):
# page = next(doc.get_pages())
#layout = page.layout() # LTPage
box_list = LeafList().populate(page, interested=flt)
ModalHeight = pdftables.calculate_modal_height(box_list)
diagnostic_data = pdftables.TableDiagnosticData(
box_list, {}, {}, [], [])
fig, ax1 = plotpage(diagnostic_data)
fig_list.append(fig)
ax1_list.append(ax1)
title = "page %d" % (i + 1)
fig.suptitle(title)
#print "Page %d" % (i+1), ElementCount
print box_list.count()
print "Modal character height: %d" % ModalHeight
return fig_list, ax1_list
def get_pdf_page(fh, pagenumber):
pdf = PDFDocument(fh)
return pdf.get_pages()[pagenumber - 1]
def get_pdf_page(fh, pagenumber):
pdf = PDFDocument(fh)
return pdf.get_pages()[pagenumber - 1]
